The uptake of Wuchereria bancrofti microfilariae (Mf) by Culex quinquefasciatus and their development in relation to human Mf density were quantified by allowing a total of 1096 wild mosquitoes to feed on 13 volunteers sleeping under partially open bed-nets. For each volunteer, each hour between 18.00 and 06.00 h the Mf density in finger-prick blood was determined and engorged mosquitoes collected. Each hourly collection of mosquitoes was kept separately. Half of them was dissected within 18 h post-feeding for the presence of ingested Mf, the other half was reared for 12 days to allow for the development of L3 larvae. About 20% of the latter mosquitoes died during these 12 days and these harboured significantly more larvae than the surviving ones, which could be an indication of excess-mortality among heavily infected mosquitoes. Assuming that variability in Mf uptake and in the number of developed L3 larvae can be described by a negative binomial distribution, a maximum-likelihood procedure was applied to estimate the relationship between human Mf density and both the arithmetic mean Mf uptake and L3 development. Both were adequately described by a saturating hyperbolic function that significantly differed from linearity. The saturation level for Mf was estimated at 29 (CI: 20–54) and for L3 larvae at 6·6 (CI: 4·3–17·0). Next, the L3 yield was related to Mf uptake indicating that the W. bancrofti–C. quinquefasciatus complex shows ‘limitation’, i.e. a decreasing yield for an increasing uptake. Both the number of Mf ingested and the number of L3 larvae developing per mosquito were found to be highly aggregated, with the level of aggregation decreasing in a non-linear way with human Mf density.